折叠 pandas 数据框中的多行，然后将数据从长数据转换为宽数据

Question

我有一个包含重复人名的数据框

输入：

{'name': {0: 'John Smith', 1: 'John Smith', 2: 'John Smith', 3: 'John Doo', 4: 'John Doo', 5: 'John Doo'}, 'journal': {0: 'Journal1', 1: 'Journal2', 2: 'Journal2', 3: 'Journal1', 4: 'Journal2', 5: 'Journal2'}, 'is_editor_in_chief_2019': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan}, 'is_editor_in_chief_2020': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan}, 'is_editor_in_chief_2021': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan}, 'is_editor_in_chief_2022': {0: 1.0, 1: nan, 2: nan, 3: 1.0, 4: nan, 5: nan}, 'is_editorial_board_member_2019': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan}, 'is_editorial_board_member_2020': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan}, 'is_editorial_board_member_2021': {0: nan, 1: nan, 2: 1.0, 3: nan, 4: nan, 5: 1.0}, 'is_editorial_board_member_2022': {0: nan, 1: 1.0, 2: nan, 3: nan, 4: 1.0, 5: nan}}

我想折叠这些行并将数据从长数据转换为宽数据。

预期输出：

{'name': {0: 'John Smith', 1: 'John Doo'}, 'Journal1_is_editor_in_chief_2019': {0: nan, 1: nan}, 'Journal1_is_editor_in_chief_2020': {0: nan, 1: nan}, 'Journal1_is_editor_in_chief_2021': {0: nan, 1: nan}, 'Journal1_is_editor_in_chief_2022': {0: 1, 1: 1}, 'Journal1_is_editorial_board_member_2019': {0: nan, 1: nan}, 'Journal1_is_editorial_board_member_2020': {0: nan, 1: nan}, 'Journal1_is_editorial_board_member_2021': {0: 1, 1: 1}, 'Journal1_is_editorial_board_member_2022': {0: 1, 1: 1}, 'Journal2_is_editor_in_chief_2019': {0: nan, 1: nan}, 'Journal2_is_editor_in_chief_2020': {0: nan, 1: nan}, 'Journal2_is_editor_in_chief_2021': {0: nan, 1: nan}, 'Journal2_is_editor_in_chief_2022': {0: nan, 1: nan}, 'Journal2_is_editorial_board_member_2019': {0: nan, 1: nan}, 'Journal2_is_editorial_board_member_2020': {0: nan, 1: nan}, 'Journal2_is_editorial_board_member_2021': {0: nan, 1: nan}, 'Journal2_is_editorial_board_member_2022': {0: nan, 1: nan}}

我找到的最接近的解决方案是 :

df_out = df.set_index(['name', df.groupby(['name']).cumcount()+1]).unstack().sort_index(level=1, axis=1)
df_out.columns = df_out.columns.map('{0[0]}_{0[1]}'.format)
df_out.reset_index()

但它没有给我理想的输出。

Answer 1

您可以使用 pivot_table:

df_out = df.pivot_table(index='name', columns='journal', dropna=False)
df_out.columns = df_out.columns.map('{0[1]}_{0[0]}'.format)
df_out.reset_index()

输出：

         name  Journal1_is_editor_in_chief_2019  Journal2_is_editor_in_chief_2019  Journal1_is_editor_in_chief_2020  Journal2_is_editor_in_chief_2020  Journal1_is_editor_in_chief_2021  Journal2_is_editor_in_chief_2021  Journal1_is_editor_in_chief_2022  Journal2_is_editor_in_chief_2022  Journal1_is_editorial_board_member_2019  Journal2_is_editorial_board_member_2019  Journal1_is_editorial_board_member_2020  Journal2_is_editorial_board_member_2020  Journal1_is_editorial_board_member_2021  Journal2_is_editorial_board_member_2021  Journal1_is_editorial_board_member_2022  Journal2_is_editorial_board_member_2022
0    John Doo                               NaN                               NaN                               NaN                               NaN                               NaN                               NaN                               1.0                               NaN                                      NaN                                      NaN                                      NaN                                      NaN                                      NaN                                      1.0                                      NaN                                      1.0
1  John Smith                               NaN                               NaN                               NaN                               NaN                               NaN                               NaN                               1.0                               NaN                                      NaN                                      NaN                                      NaN                                      NaN                                      NaN                                      1.0                                      NaN                                      1.0

折叠 pandas 数据框中的多行，然后将数据从长数据转换为宽数据

Collapse multiple rows in pandas dataframe and convert data from long to wide afterwards

python

collapse

pandas